espnet3.systems.base.system.BaseSystem

source

class espnet3.systems.base.system.BaseSystem(, train_config: DictConfig | None = None, infer_config: DictConfig | None = None, metric_config: DictConfig | None = None, publish_config: DictConfig | None = None, demo_config: DictConfig | None = None, demo_config_path: Path | None = None)

Bases: object

Base class for all ESPnet3 systems.

Each system should implement the following: : - create_dataset()

• train()
• infer()
• metric()
• publish()

This class intentionally does NOT implement: : - DAG

• dependency checks
• caching

All behavior is config-driven.

Initialize the system with optional stage configs.

collect_stats(*args, **kwargs)

Collect statistics needed for training.

create_dataset(*args, **kwargs)

Create datasets using the configured stage.

get_stage_log_dir(stage: str) → Path

Return the default log directory for a stage.

BaseSystem treats all stages uniformly and writes logs into: : - train_config.exp_dir when available, or

• <cwd>/logs as a fallback when no experiment directory is set.

Subclasses can override this method to route specific stages to stage-aware artifact locations (e.g., dataset or decode outputs).

• Parameters:stage (str) – Stage name being executed (unused by BaseSystem).
• Returns: Directory where the stage log should be placed.
• Return type: Path

infer(*args, **kwargs)

Run inference on the configured datasets.

metric(*args, **kwargs)

Compute evaluation metrics from hypothesis/reference outputs.

pack_demo(*args, **kwargs)

Pack demo artifacts into a runnable demo bundle.

pack_model(*args, **kwargs)

Pack model artifacts into an espnet3 bundle.

publish(*args, **kwargs)

Publish artifacts from the experiment.

train(*args, **kwargs)

Train the system model.

upload_demo(*args, **kwargs)

Upload demo bundle to HuggingFace Spaces (stub).

upload_model(*args, **kwargs)

Upload model bundle to HuggingFace.